Oracle 18c 备库报错ORA-01110， Failed to request gap sequence

备库报错，日志如下：

2019-08-08T09:46:32.915070+02:00 
PR00 (PID:10646): Media Recovery Waiting for T-1.S-2306 
PR00 (PID:10646): Fetching gap from T-1.S-2306 to T-1.S-2405 
2019-08-08T09:46:32.925843+02:00 
Errors in file /u01/app/oracle/diag/rdbms/xxxp2/XXXP2/trace/XXXP2_mz00_10650.trc: 
ORA-01110: data file 1: '+ORADATA/XXXP2/DATAFILE/system.263.1009981381'

2019-08-08T09:48:27.919844+02:00 
PR00 (PID:10646): FAL: Failed to request gap sequence 
PR00 (PID:10646): GAP - thread 1 sequence 2306-2405 
PR00 (PID:10646): DBID 123456789 branch 994167271 
PR00 (PID:10646): FAL: All defined FAL servers have been attempted 
PR00 (PID:10646): ------------------------------------------------------------------------- 
PR00 (PID:10646): Check that the CONTROL_FILE_RECORD_KEEP_TIME initialization 
PR00 (PID:10646): parameter is defined to a value that's sufficiently large 
PR00 (PID:10646): enough to maintain adequate log switch information to resolve 
PR00 (PID:10646): archived redo log gaps. 
PR00 (PID:1064

18c bug,当应用进程启用之后，数据文件的CHECKPOINT未更新。
BUG 29056767 – DATAFILES CHECKPOINT NOT UPDATED AT STANDBY DATABASE WHEN MRP IS ENABLED

这个问题只发生在18.1及以上版本，当备库MRP进程恢复在线日志时，数据文件的checkpoint scn and time没有更新，

  select to_char(sysdate,'HH24:MI:SS'), file#, checkpoint_change#,
      to_char(checkpoint_time,'HH24:MI:SS') from v$datafile_header;

这个bug的影响：

• level 1增量备份会跳过所有数据文件。
• 如果mrp突然中止，比如shutdown abort，那么下次启动恢复进程是就会去找redo，如果redo不存在，那么就会出现gap报错。

临时解决方案:
在备库设置_time_based_rcv_ckpt_target参数并重启恢复进程:

alter system set "_time_based_rcv_ckpt_target"=0;
--ALTER DATABASE RECOVER MANAGED STANDBY DATABASE CANCEL;
--ALTER DATABASE RECOVER MANAGED STANDBY DATABASE DISCONNECT FROM SESSION;
--alter database recover managed standby database using current logfile disconnect from session;

另外建议主库也设置下参数，以防以后主备切换出现同样的错误。